Package Details: catalyst-total-hd234k 13.1-42

Git Clone URL: (read-only)
Package Base: catalyst-total-hd234k
Description: AMD/ATI legacy drivers. catalyst-dkms+ catalyst-utils + lib32-catalyst-utils
Upstream URL:
Licenses: custom
Conflicts: catalyst, catalyst-daemon, catalyst-dkms, catalyst-generator, catalyst-test, catalyst-utils, libegl, libgl, libgles, libglvnd, mesa-libgl, mesa-libgl-git, nvidia, nvidia-utils, opencl-amd, xf86-video-ati, xf86-video-radeonhd
Provides: catalyst=13.1, catalyst-dkms=13.1, catalyst-libgl=13.1, catalyst-utils=13.1, dri, libatical=13.1, libegl, libgl, libgles, libtxc_dxtn, opencl-catalyst=13.1, opencl-driver
Submitter: Vi0L0
Maintainer: Vi0L0
Last Packager: Vi0L0
Votes: 33
Popularity: 0.017206
First Submitted: 2012-06-12 16:59
Last Updated: 2017-04-03 18:59

Dependencies (23)

Required by (631)

Sources (35)

Latest Comments

Thormme commented on 2017-10-11 06:45

I had the exact same problem with the same log output as AwesomeHaircut.
It's not a good solution, but I have a workaround:
Downgrade to a pre-4.8 linux/linux-headers, I used 4.6.5-1
At this point the fglrx driver won't rebuild because of pie defaulting to on by default in gcc 6.
Downgrade to gcc/gcc-libs 5.3
Rebuild fglrx (catalyst-hook)

At this point startx works for me, though the outdated GLIBCXX causes a number of programs to fail. If you managed to disable pie for the fglrx rebuild you could probably get it working with a newer gcc.

Vi0L0 commented on 2017-04-08 18:29

Thanks AwesomeHaircut.
Part at below seems to show the problem. For now I don't know how to bite it, but it brings more info than I ever had on this topic. Will make some research, any help will be appriciated.

315.932849] <6>[fglrx] Reserved FB block: Unshared
offset:fbfa000, size:401000
[ 315.932850] <6>[fglrx] Reserved FB block: Unshared
offset:fffb000, size:5000
[ 315.945474] usercopy: kernel memory overwrite attempt detected
to ffffc90003363d7c (<process stack>) (4 bytes)
[ 315.945505] ------------[ cut here ]------------
[ 315.945506] kernel BUG at mm/usercopy.c:75!
[ 315.945509] invalid opcode: 0000 [#1] SMP
[ 315.945510] Modules linked in: ctr ccm xt_REDIRECT
nf_nat_redirect xt_tcpudp nf_nat_pptp nf_nat_proto_gre
nf_conntrack_pptp nf_conntrack_proto_gre iptable_filter
ipt_MASQUERADE nf_nat_masquerade_ipv4 iptable_nat
nf_conntrack_ipv4 nf_defrag_ipv4 nf_nat_ipv4 nf_nat
nf_conntrack crc32c_generic btrfs xor raid6_pq mousedev joydev
hid_uclogic hid_generic edac_mce_amd edac_core kvm_amd kvm
acer_wmi sparse_keymap arc4 irqbypass psmouse input_leds pcspkr
ath9k_htc k10temp ath9k ath9k_common ath9k_hw ttm usbhid
snd_hda_codec_realtek hid led_class drm_kms_helper ath
mac80211 snd_hda_codec_generic snd_hda_codec_hdmi drm
broadcom snd_hda_intel snd_hda_codec bcm_phy_lib snd_hda_core
sp5100_tco syscopyarea sysfillrect sysimgblt fb_sys_fops cfg80211
snd_hwdep tg3 i2c_piix4 i2c_algo_bit rfkill snd_pcm ptp
[ 315.945539] snd pps_core libphy soundcore shpchp acpi_cpufreq
thermal tpm_tis tpm_tis_core fjes tpm wmi evdev video battery
mac_hid ac sch_fq_codel msr videodev media sg ip_tables x_tablesext4 crc16 jbd2 fscrypto mbcache sd_mod ohci_pci serio_raw atkbd
libps2 ahci libahci libata ohci_hcd ehci_pci ehci_hcd usbcore
usb_common scsi_mod i8042 serio fglrx(O) button amd_iommu_v2
[ 315.945559] CPU: 1 PID: 3097 Comm: X Tainted: G O
4.9.20-1-lts #1
[ 315.945560] Hardware name: Acer Aspire 5251
JE50_DN, BIOS V2.14 07/27/2011
[ 315.945562] task: ffff8801efdae800 task.stack: ffffc90003360000
[ 315.945563] RIP: 0010:[<ffffffff8120555f>] [<ffffffff8120555f>]
[ 315.945569] RSP: 0018:ffffc90003363d18 EFLAGS: 00010286
[ 315.945570] RAX: 0000000000000062 RBX: ffffc90003363d7c RCX:
[ 315.945572] RDX: 0000000000000000 RSI: ffff88020fc8dc68 RDI:
[ 315.945573] RBP: ffffc90003363d38 R08: 00000000000e6d42 R09:
[ 315.945574] R10: 00000000000000c7 R11: 0000000000000326 R12:
[ 315.945575] R13: 0000000000000000 R14: ffffc90003363d80 R15:
[ 315.945577] FS: 00007ffb2dee1fc0(0000)
GS:ffff88020fc80000(0000) knlGS:0000000000000000
[ 315.945578] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 315.945579] CR2: 00007ffb2a027720 CR3: 00000001f7ea9000 CR4:
[ 315.945580] Stack:
[ 315.945582] ffffc90003363d7c 0000000000000004
00007fffb04fa584 000000004004645a
[ 315.945584] ffffc90003363d60 ffffffffa0015e41 00007fffb04fa584
[ 315.945586] ffff8801f7eb4000 ffff8802046bc000 ffffffffa0025fb6
[ 315.945588] Call Trace:
[ 315.945664] [<ffffffffa0015e41>] KCL_CopyFromUserSpac
+0x21/0x40 [fglrx]
[ 315.945690] [<ffffffffa0025fb6>] firegl_set_block_mod
+0x56/0xb0 [fglrx]
[ 315.945694] [<ffffffff81179854>] ? unlock_page+0x34/0x40
[ 315.945697] [<ffffffff812b1c78>] ? security_capable+0x48/0x60
[ 315.945700] [<ffffffff810868b8>] ? ns_capable_commo
[ 315.945701] [<ffffffff8108690c>] ? capable+0x1c/0x20
[ 315.945727] [<ffffffffa0025f60>] ? __ke_strcpy+0x10/0x10
[ 315.945753] [<ffffffffa002416d>] ? firegl_ioctl+0x1ed/0x250
[fglrx][ 315.945776] [<ffffffffa001337e>] ? ip_firegl_unlocked_ioct
+0xe/0x20 [fglrx]
[ 315.945778] [<ffffffff8121b853>] ? do_vfs_ioctl+0xa3/0x5f0
[ 315.945781] [<ffffffff81067008>] ? __do_page_faul
[ 315.945783] [<ffffffff8121be19>] ? SyS_ioctl+0x79/0x90
[ 315.945786] [<ffffffff815f9a37>] ? entry_SYSCALL_64_fastpat
[ 315.945787] Code: bd 91 81 48 0f 45 d0 48 c7 c6 f7 e0 92 81 48 c7
c0 e3 81 93 81 48 0f 45 f0 4d 89 e1 48 89 d9 48 c7 c7 e0 49 93 81 e8
26 31 f7 ff <0f> 0b 48 89 df e8 47 7a e6 ff 84 c0 0f 84 f8 fe ff ff b8
00 00
[ 315.945807] RIP [<ffffffff8120555f>] __check_object_siz
[ 315.945809] RSP <ffffc90003363d18>
[ 315.945812] ---[ end trace 83d63bbd701ebdc7 ]---
[ 315.946636] <3>[fglrx:firegl_release] *ERROR* device busy: 1
[ 315.946639] <6>[fglrx] release failed with code -EBUSY

AwesomeHaircut commented on 2017-04-08 00:07

You'd think after doing this so many times I could do it without the wiki but nope. Still a bit much.
The driver behaves the same way with Linux 4.10 and Linux-LTS 4.9.

I haven't gotten this driver working ever though so idk.


Hey thanks a lot for the work Vi0L0, really appreciate it! :D

Kutabe commented on 2017-04-07 01:16

I have the same issue as AwesomeHaircut with catalyst-hd234k
I've installed packages from repo
Here is what I've done
Xorg.0.log is empty, I'll add
- journalctl -b
- dmesg

Vi0L0 commented on 2017-04-03 19:03

If you won't have luck could you paste on some pastebin?:
- Xorg.0.log
- try to run startx and note what it returns
- dmesg
- journalctl -b

AwesomeHaircut commented on 2017-04-03 03:40

I got to where it was all installed and configured but when I started X it gave me a black screen and unresponsive keyboard. ssh worked though so that saved some reboots. I ended up going back to the open source driver for work but I do plan to give it another shot tomorrow.

I'll try out those other packages thanks!

Vi0L0 commented on 2017-04-02 16:41

@AwesomeHaircut: also did this package work for you?
I'm asking because I don't have confirmation.
Also please note that to run it in post-libglvnd era you have to use pre-libglvnd mesa's packages, you can find it ie in [catalyst] repo.
I am planning to also paste it in [catalyst-hd234k] repo
Edit: I also created non-libglvnd packages: mesa-noglvnd and lib32-mesa-noglvnd groups
(tho on AUR it comes with *mesa-libgl* packages which should be avoid by catalyst users, I left it there for package base to be more "general")

Vi0L0 commented on 2017-04-02 16:38

@AwesomeHaircut: you are right, thanks.
I don't know where this *** came from...
Anyhow should be fine now

AwesomeHaircut commented on 2017-03-30 13:24

I think you forgot to add "hd234k" (or some other change) at the end of these two changes in build() in your last commit :

+ patch -Np1 -i ../4.9_over_4.6-arch-get_user_pages_remote.patch
+ patch -Np1 -i ../4.10-arch-sling00-virtual_address-acpi_get_table_with_size.patch

I kept getting file missing when using makepkg -s

patch: **** Can't open patch file ../4.9_over_4.6-arch-get_user_pages_remote.patch : No such file or directory

I have to append hd234k so that it can build.

Vi0L0 commented on 2017-03-04 18:14

I was able to succesfully compile module on 4.10.1 kernel with this new patch.
Can't promise if it will really work, I'm afraid that it needs general rewrite and not some simple patches...

All comments